Credit Rating Models And Validation Of Commercial Bank

Along with economic globalization, the gradual increase in the openness of the banking industry, how to guard against financial risks is a major issue faced by China's banking industry. In the risks faced by commercial banks, the credit risk through the commercial bank's operations from the beginning to the end, which highlights the importance of credit risk management. This paper presents a credit risk management models-probability of default (PD) model.Internationally, it has formed two kinds of basic methods for the choice of variables of the default probability model. The first method, commonly referred to as stepwise regression. It uses a convenient calculation algorithm, limits the number of possible models to a very small number. Although it chooses a model does not correspond to any particular criteria, but it was frequently used in practice. The second method uses the criteria for statistics calculated on all the possible subset of variables. It is all possible with the calculation of the return of several fast algorithm development, and will gradually be applied to practice. In this paper, a new variable selection method, based mainly on the correlation of variables and the dependent variable,has been used. It calculates the variable's Weight Of Evidence (WOE) and the Information Value (Ⅳ). This article also highlights the most commonly used in practice, the Logistic Regression Model. And we evaluate the pros and cons of the model by constructing the ROC curve and the CAP curve.In Schweizcr & WolR(1981).'Copula' was taken as keyword the first time in history, which was translated in chinese as "关联函数" and was used to measure the dependence of random variables.This article concludes with a commercial bank for the establishment of an industry credit rating model, shows that our model is reasonable.